produced content that may be false or of low quality. Deepfakes One of the most prominent forms of misuse of generative AI technology is deepfakes. Deepfakes are synthetic media that depict a person as saying or doing things they did not actually say or do, and they carry particularly high risks of being used for violence or fraud, including sexual exploitation targeting individuals. In South Korea, organized deepfake-based sexual crimes have already emerged as a serious social problem. While such crimes existed even before the advent of generative AI technologies, generative AI facilitates them. As a result, these technologies raise new challenges across multiple areas, including criminal punishment, prevention, technical countermeasures, and victim support and recovery. Security and Privacy The performance of generative AI models has tended to improve as the volume of training data and the size of the model—often expressed in terms of the number of parameters—increase. As a result, the generative AI industry has pursued the collection of as much data as possible, often at the expense of careful attention to the legality and quality of that data. The practice of collecting publicly available personal data online for use in building generative AI systems raises serious privacy concerns and may conflict with core data protection principles, such as data minimization. Moreover, information collected in this manner may later be exposed 38 39